BigTop: a three-dimensional virtual reality tool for GWAS …helios.mi.parisdescartes.fr/~lomn/Cours/BI/Articles/10a... · 2020. 3. 18. · SOFTWARE Open Access BigTop: a three-dimensional

Westreich et al. BMC Bioinformatics (2020) 21:39 https://doi.org/10.1186/s12859-020-3373-5

SOFTWARE Open Access

BigTop: a three-dimensional virtual reality

Abstract

Background: Genome-wide association studies (GWAS) are typically visualized using a two-dimensional Manhattanplot, displaying chromosomal location of SNPs along the x-axis and the negative log-10 of their p-value on the y-axis. This traditional plot provides a broad overview of the results, but offers little opportunity for interaction orexpansion of specific regions, and is unable to show additional dimensions of the dataset.

Results: We created BigTop, a visualization framework in virtual reality (VR), designed to render a Manhattan plot inthree dimensions, wrapping the graph around the user in a simulated cylindrical room. BigTop uses the z-axis todisplay minor allele frequency of each SNP, allowing for the identification of allelic variants of genes. BigTop alsooffers additional interactivity, allowing users to select any individual SNP and receive expanded information,including SNP name, exact values, and gene location, if applicable. BigTop is built in JavaScript using the React andA-Frame frameworks, and can be rendered using commercially available VR headsets or in a two-dimensional webbrowser such as Google Chrome. Data is read into BigTop in JSON format, and can be provided as either JSON or atab-separated text file.

Conclusions: Using additional dimensions and interactivity options offered through VR, we provide a new,interactive, three-dimensional representation of the traditional Manhattan plot for displaying and exploring GWASdata.

Keywords: Virtual reality, User Interface, Visualization, GWAS, Manhattan plot, Data plotting, Data interaction

BackgroundIn the last two decades, a decrease in the cost of sequen-cing has led to a steep increase in the amount of geneticinformation generated. One aspect of this proliferationof data is an increase in genome-wide association studies(GWAS), each of which requires thousands of individ-uals to be genotyped or sequenced. In order to interpretthe results of a GWAS, it is necessary to condense thelarge amount of information into a graphic that is stillreadable and understandable.The classic visualization for GWAS results is the Man-

hattan plot [1]. Named because of its resemblance to theskyline of a city with a row of tall buildings, the Manhat-tan plot shows associations for variants across the gen-ome with a given phenotype. Each point displayed on aManhattan plot represents a single point mutation, or

© The Author(s). 2020 Open Access This articInternational License (http://creativecommonsreproduction in any medium, provided you gthe Creative Commons license, and indicate if(http://creativecommons.org/publicdomain/ze

* Correspondence: [email protected]†Samuel T. Westreich, Maria Nattestad and Christopher Meyer contributedequally to this work.DNAnexus, Inc., Mountain View, CA, USA

single nucleotide polymorphism (SNP), with thechromosome position plotted along the X axis, and thenegative log of the P value for the association test shownon the Y axis. While most measured SNPs have lownegative log P values indicating that their associations tothe trait being measured by the GWAS are not signifi-cant, some SNPs will be highly associated and will thusappear higher on the Y axis [2].The typical Manhattan plot is useful for providing an

overview of the GWAS, showing where significant asso-ciations exist on a whole-genome view. These plots arerendered either as static images or as interactive visuali-zations [3–5]. However, a typical two-dimensional Man-hattan plot has several drawbacks inherent to itsmedium: 1) the density of information can potentiallyobscure interesting results; 2) even in interactive Man-hattan plots, selecting a point of interest can be difficultwithin a dense cluster; 3) additional context such as thepopulation-level allele frequency could aid with inter-pretation of the results. In addition, standard

le is distributed under the terms of the Creative Commons Attribution 4.0.org/licenses/by/4.0/), which permits unrestricted use, distribution, andive appropriate credit to the original author(s) and the source, provide a link tochanges were made. The Creative Commons Public Domain Dedication waiverro/1.0/) applies to the data made available in this article, unless otherwise stated.

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 2 of 8

visualization methods for adding dimensionality (such asvaried colors, textures, or shapes) will not work due tothe density of information, meaning that adding extracontext to a two-dimensional Manhattan plot presentsas difficult to impossible.Of course, traditional static Manhattan plots also lack

the ability to zoom in to observe details about specificSNPs, and generally do not provide any identification ofindividual SNPs unless this information is manuallyoverlaid on the figure through image editing software.Static Manhattan plots also fail to offer additional infor-mation about specific SNPs, such as relative abundanceor specific chromosomal position. Interactive Manhattanplots offer improvement in many of these areas, butsome problems persist due to the natural limitations oftwo dimensions.An innovation in technology that is being applied to

large genomic datasets is virtual reality (VR). VR applica-tions have been created for several subfields of biologyand genetics, including visualization of synteny [6], tra-cing of neural pathways [7], or three-dimensional pro-tein structure [8–10]. These visualizations exist nativelyin a three-dimensional environment, making them idealcandidates for exploration in virtual reality.VR is ideal for visualizing large amounts of data that

may not be suitable for the constrained display space oftwo-dimensional monitors. It also permits interaction,allowing for the exploration of data within the figure byobservers. A VR-based framework for visualization ofgenetic or genomic data should be flexible, allowing vari-ous datasets to be imported and rendered without re-quiring any modification of the source code.One drawback to VR-based visualizations is that cre-

ation of the visualization requires a combination of mul-tiple skills. These visualizations are typically createdusing either WebXR in HTML [6] or an applicationframework such as Unity or Unreal Engine [8, 10], re-quiring considerable programming experience. Addition-ally, VR equipment is not yet widely deployed, meaningits availability to researchers may be limited. An idealVR visualization application should 1) require minimaltechnical expertise on the user’s part, and 2) be able todisplay information in a virtual world using a standardmonitor.We created BigTop, a React-based [11] web applica-

tion that uses the A-Frame framework [12] to render in-put GWAS summary data in three dimensions. BigToplaunches an interactive three-dimensional environmentthat renders GWAS summary data in three dimensions,wrapping the data in a cylindrical fashion around theuser similar to other cylindrical visualizations such asCircos [13]. BigTop supports data interaction eitherthrough a VR headset or through the combination of amonitor, mouse, and keyboard, allowing users to

navigate within the environment and select individualdata points to glean more information. Data is read intoBigTop in JSON format, but can be provided as a multi-column TSV file and converted to JSON by an includedscript.

ImplementationSystem overview

1. GWAS data is provided in JSON format, specifyingthe chromosome, SNP location on thechromosome, negative log-ten of the p-value, andanother measurement used in the z-axis (in all ex-amples, minor allele frequency is used)

Additionally, a separate preprocessing script canbe run on the data to provide information aboutthat SNP’s location (gene name if in a gene).This script only needs to be run once per inputfile.

2. For non-human data, a separate file containschromosome number and size. This can be re-placed by the chromosome count and sizes forany other organism.

2. BigTop is easily installed and runs on any systemwith JavaScript. The display loads in the Chrome orFirefox browsers, and can be viewed through a VRheadset with the click of a button. BigTop has beentested and performs on the Oculus Rift, the HTCVive, and the smartphone-based Google Daydream.

3. BigTop wraps the traditional Manhattan plotaround a cylindrical room, placing the user in thecenter of the room. Chromosomes are marked andcolored on the walls, while the height of each pointcorresponds to the negative log-10 p-value, and thedistance along the z-axis (from the center of theroom to the wall) indicates the third measurement(minor allele frequency in all example data) (Fig. 1).

4. The user can move around the room by takingsteps (with a VR headset) or by using the arrowkeys (if using a browser). They may control wherethey look by either moving their head (VR) or byusing the mouse to click and drag (browser).

5. In VR, one of the hand controls is set to be a laserpointer (the hand may be switched in BigTopsettings). Aiming this laser pointer at a point andpulling the trigger to select that point will displayan info panel near the point, providing additionalinformation such as exact p-value, SNP location,gene name, and SNP name (if using human data),and more.

reference beams to the floor and far wall, better

Fig. 1 Illustration of the mapping of SNPs in a three-dimensional space. Chromosomal position is indicated by the circumferential locationaround the edge of the room, while the height of the point indicates the negative log-10 p-value, and the radial distance from the centerindicates allele frequency

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 3 of 8

allowing the user to gauge where it falls on thedifferent axes.

6. If using a browser, point selection is possible bycentering the point in the center of the user’svision. A targeting reticule helps align the camerawith a point of interest.

Hardware requirementsBigTop was tested and developed in non-VR mode on a2018 MacBook Pro with the following specifications: (1)3.1 GHz Intel Core i5, (2) 16 GB 2133 MHz LPDDR3RAM, (3) 512GB SSD, (4) Intel Iris Plus Graphics 6,501,536MB graphics card. BigTop was also tested in VRmode on a laptop computer with the following hard-ware: (1) Intel i7-7700HQ CPU, (2) 16GB DDR4 RAM,(3) 256GB SSD, (4) GeForce GTX 1060-6GB graphicscard. An Oculus Rift was used, with cameras on firm-ware version 178/e9c7e04064ed1bd7a089, headset onfirmware version 709/b1ae4f61ae.

Development toolsBigTop is coded entirely in JavaScript using the Reactframework, and makes use of the primitives provided inthe A-Frame framework for three-dimensional render-ing. A-Frame [12] and other components are installedvia the npm package manager. From the Chrome orFirefox web browsers, BigTop may be launched forOculus Rift, HTC Vive, or another VR display.

Data Structure & ImportBigTop accepts input data in a structured JSON format.At minimum, four fields must be provided for each SNP

- chromosome number, chromosome position, minor al-lele frequency (MAF), and p-value. Additional data, suchas the SNP name and/or the gene location of the SNP,may also be provided and are displayed in the informa-tion panel for any selected gene within the visualization.To aid in preparing data for use with Bigtop, a Python

script, SNP_info_retriever, provided with BigTop canconvert a tab-separated values (TSV) file with the re-quired information to JSON format. A tab-separatedvalues spreadsheet can be structured with the columnscontaining, in order, the rsID, chromosome number,major allele, minor allele, MAF and p-value. Based uponthe SNP rsID, information such as the chromosomal andgene location are retrieved from the UCSC GenomeBrowser using the cruzdb plugin [14]. If working withGWAS data from a non-human source, the nonhuman_JSON_setup.py script can convert a TSV spreadsheetcontaining just chromosome number, chromosome loca-tion, MAF, and p-value for each SNP to structuredJSON format.To dynamically render the BigTop environment and

background, chromosome count and lengths are pro-vided in a chrInfo.json file, and cytoband information isprovided in a cytobands.json file. Chromosome count,lengths, and cytoband information is provided for hu-man and rice data.

Data rendering & UIBased on information in the chrInfo and cytobands files,BigTop renders the organism genome as a three-dimensional cylindrical room. Each chromosome is rep-resented by a pie slice of the room, and the cytobands

Fig. 2 initial view of BigTop upon rendering. The camera loads facing chromosome 1, in the center of the defined “room”. The camera may berotated to view other areas, either by using the mouse or by turning while wearing an active VR headset

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 4 of 8

are on each chromosome at approximately eye height.Horizontal and vertical axes are rendered on the floorand on the walls at the intersection between eachchromosome.Input data is used to render each SNP as an object within

the VR environment. To improve computational perform-ance, reduce lag, and improve clarity, only SNPs with nega-tive log-10 p-value above a set threshold are rendered asthree-dimensional spheres. SNPs with a negative log-10 p-value below the cutoff threshold are instead rendered as cir-cular shadows that are mapped to the floor. The height axisis automatically scaled to reflect the minimum and max-imum negative log-10 p-values that are rendered.Activating a specific SNP by selecting the sphere, ei-

ther using hand controls in VR or with the cursor, dis-plays additional information about the selected point. Asphere becomes marked as active after being selected inthis manner, and an informational panel appears adja-cent to the active sphere. This informational panel dis-plays the exact location and p-value of the selected SNP,along with additional information, such as SNP nameand gene location, if provided in the input file. Addition-ally, guides extend from the active sphere to the floorand wall axes to better mark its position in three-dimensional space.

Within the rendered virtual environment, user move-ment may be performed using arrow keys, if viewing in abrowser, or by physically moving while wearing a VRheadset.

Customizing the InterfaceThe display of data can be customized by altering theURL. All available parameters are described in the docu-mentation at https://github.com/dnanexus/bigtop#confi-guration, and include the ability to set a maximum p-value threshold, highlight specific genes or rsIDs, or de-fine the maximum number of points to render for per-formance purposes. URL parameters can also be alteredto customize interface elements, such as switching to aleft-handed interface, customizing the perceived heightand radius of the room, or displaying a list of perform-ance statistics on the screen.

Test datasetsBigTop is capable of displaying GWAS summary datafrom any organism, as long as the four default values areprovided per SNP (chromosome, location, MAF, and p-value). For any organism, chromosome number and sizesmust be provided in the chrInfo.json file. BigTop is dis-tributed with two human GWAS datasets; a GIANT

Fig. 3 view of an info panel, with a selected SNP. The info panel always displays negative log-10 p-value, chromosome number, andchromosome position, as well as other values (such as SNP name and gene location) if provided in the input file

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 5 of 8

dataset which examines SNPs correlated with humanheight [15], and a breast cancer GWAS from the Nurses’Health Study, dbGaP accession phs000147.v3.p1 [16,17]. To demonstrate its use across multiple species, Big-Top also includes a dataset from Oryza sativa, examin-ing SNPs related to grain size [18]. Toggling betweendatasets may be accomplished by switching the informa-tion in chrInfo.json and cytobands.json, and changingthe data file referenced in the main app.js script.

ResultsExploring Human & non-Human DataThe BigTop rendered 3D image allows a user to explorehis or her data through interaction; initially loading inthe center of the cylindrical image of a GWAS wrappedaround the user, the user may move both the locationand angle of the camera. As seen in Fig. 2, the cameraloads facing the start of chromosome 1; in order to viewall the data, the user must pivot the camera angle.While the BigTop VR simulation is running, the user

may interact with any specific data point by selecting it.If using BigTop with a VR headset and hand controls, alaser pointer is attached to one hand. This laser pointercan be aimed at any individual SNP, represented as asphere, and an information panel will appear when the

trigger is pulled (Fig. 3). If using BigTop with a webbrowser, a small circle in the center of the screen acts asa heads-up display; centering this circle on any individ-ual sphere will bring up the information panel.Exploration of GWAS data in 3-dimensional space al-

lows for new insights, such as the distribution of SNPsby minor allele frequency. Where a two-dimensionaltraditional Manhattan plot would show a spike of SNPsat a significant locus, BigTop allows the user to observethe distribution of these SNPs based on their minor al-lele frequency, or another variable that can be measuredand plotted on the Z axis (Fig. 4). This additional dimen-sion of information also allows the user to select specificSNPs for further investigation with the consideration oftheir relative frequency as an additional weight.

Installation-free rendering from a URLA downside of most VR applications is that they requirethe user to install and set up an environment on theirlocal machine before he or she may view and explorethe VR world. This limitation prevents the VR displayfrom being widely shared or distributed. BigTop circum-vents this limitation by allowing users to run the tool byvisiting a web page using a browser that supports VR(such as properly-configured installations of Chrome or

Fig. 4 Clusters of significant variants at different frequencies. Multiple variants of significant association were found at clusters of different allelefrequencies, suggesting clusters of variants carried together through linkage disequilibrium. This pattern is only visible in three dimensions (right,fig. 4b) and not when looking straight down the axis (left, fig. 4a) as one would in a regular 2-dimensional Manhattan plot. Numbers on thepanels in the figure indicate the chromosome(s) being viewed

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 6 of 8

Firefox). An example of BigTop can be viewed here:https://dnanexus.github.io/bigtop/build/ .This method also allows BigTop results to be easily in-

corporated into external publications, such as blog postsor scientific papers. The increase in popularity of openaccess publications and methods has led to an increasingnumber of researchers making their scripts and methodsavailable on sites such as GitHub. A BigTop visualizationcan easily be included on such an external site, allowingthe researchers to include this interactive figure in theirpublication.

All configuration in BigTop is done through URL pa-rameters, making it easy to change data sets andcustomize the user’s experience. Full documentation forgetting started can be found at https://github.com/dna-nexus/bigtop, but in order to use data sets other thanthe defaults, a user would visit a URL similar to https://dnanexus.github.io/bigtop/build/?data=[dataURL], re-placing “[dataURL]” with the URL to the data file, instructured JSON format, to be visualized.Data files are assumed to be linked to GRCh38, but

BigTop will accept any other chromosome and cytoband

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 7 of 8

files, permitting customization of the background andenvironment to include any number of chromosomes ofany chosen size, with any chosen pattern of displayedcytobands. These can also be specified in the URL, such ashttps://dnanexus.github.io/bigtop/build/?data=[dataURL]&chr=[chrURL]&cyto=[cytoURL], where “[dataURL]”,“[chrURL]”, and “[cytoURL]” all refer to the URLs of therespective files. Further information on the formats ofthese files can be found in the documentation.

DiscussionAdvances in genetic sequencing and analysis technologyhave led to a greater emphasis on bioinformatics tools forexploring large datasets. Experimental data is now oftentoo large to manually curate and requires specialized soft-ware to view and interpret. Presenting high-density visual-izations while still providing information in an easilyinterpreted schema creates additional challenges.In this paper, we present BigTop, an easy-to-use, inter-

active VR-based method for exploring genome-wide as-sociation study (GWAS) data. This tool aims to increasethe information density of the traditional GWAS Man-hattan plot, while also allowing interactivity and in-creased customization and ease of multi-dimensionalexploration. BigTop can be run on a local machine orwith multiple currently available VR platforms, includingOculus Rift, HTC Vive, and Google Daydream. BigToplaunches as an HTML object in a Chrome web browseron Windows, Mac, and Unix machines, and can behosted on an external website, such as GitHub Pages, toallow other users to access and explore the visualizationwith a link without needing to install or setup any com-ponents on their local machine. BigTop reads in a sim-ple comma-separated list, and requires only fourcomponents per SNP to render it in three dimensions.BigTop can handle GWAS data from any organism, andsettings such as the number and size of chromosomesand pattern of cytoband staining may be altered to suitthe chosen organism.BigTop has several limitations, and there is room for

continued improvement and expansion. Currently, Big-Top does not support swapping datasets without theneed for direct editing of the source code, or by pointingto the new dataset via URL. In a future update, we lookto bring a graphical user interface (GUI) to the tool,allowing for the selection of the input dataset, chromo-some information, and cytoband pattern information.Additional planned improvements include addinggreater interactivity within the visualization using VRcontrols.

ConclusionsAdvances in the ease and rapidity of gathering GWASdata has made it easier and more affordable than ever

before to perform a GWAS to investigate a trait of inter-est or as part of a larger study. GWAS data is usually vi-sualized in a two-dimensional Manhattan plot, showingthe relation between SNP location on the genome andits p-value, indicating how likely it is to be associatedwith the trait of interest. BigTop allows for visualizationof GWAS summary data in three dimensions, includingminor allele frequency (MAF) as an additional axis, andallows users to interactively query any individual pointfor more information. It also includes filters and can beused for GWAS data from any custom organism orsource.

Availability and requirementsProject Name: BigTop.Project Home Page: https://github.com/dnanexus/bigtopOperating System(s): Platform independent.Programming Language: JavaScript.Other Requirements: npm, WebXR-enabled browser

(either Google Chrome or Mozilla Firefox).License: MIT license.Restrictions: No restrictions.

AbbreviationsGWAS: Genome-wide association study; JSON: JavaScript Object Notation;MAF: Minor allele frequency; SNP: Single nucleotide polymorphism; TSV: Tab-separated values; VR: Virtual reality

AcknowledgementsThe authors wish to thank their colleagues at DNAnexus who aided in thedevelopment and testing of the BigTop framework, including AndrewCarroll, who provided funds for the purchase of testing equipment.

Authors’ contributionsInitial design of BigTop provided by SW, MN, and CM. Code was written anddeveloped by CM and MN. Data was obtained and structured by SW. Allauthors contributed to and approved the final manuscript.

FundingAll authors were employed by DNAnexus, Inc. during the creation of thesoftware, and DNAnexus, Inc. provided the authors’ salaries during thedevelopment of the software. All data was publicly sourced and collectedfrom public repositories.

Availability of data and materialsAll components of BigTop are available publicly on GitHub at https://github.com/dnanexus/bigtop. The repository includes documentation andinstructions for installation and running the program. The repositoryadditionally includes public data that can be rendered.All starting data used in development and implementation of BigTop arepublicly available. Summary datasets used by BigTop for visualization areavailable with the source code in the public repository.

Ethics approval and consent to participateNot applicable.

Consent for publicationNot applicable.

Competing interestsAll authors received salaries from DNAnexus, Inc. during the creation of thesoftware.

Westreich et al. BMC Bioinformatics (2020) 21:39 Page 8 of 8

Received: 17 May 2019 Accepted: 17 January 2020

References1. Dubé JB, Hegele RA. Genetics 100 for cardiologists: basics of genome-wide

association studies. Can J Cardiol. 2013;29:10–7.2. Visscher PM, Wray NR, Zhang Q, Sklar P, McCarthy MI, Brown MA, Yang J. 10

Years of GWAS Discovery: Biology, Function, and Translation. Am J HumGenet. 2017;101:5–22.

3. Sahir Bhatnagar (2016). manhattanly: Interactive Q-Q and Manhattan PlotsUsing 'plotly.js'. R package version 0.2.0. https://CRAN.Rproject.org/package=manhattanly

4. Barrios D, Prieto C. RJSPlot: Interactive Graphs with R. Mol. Inf. 2018;37:1700090.

5. Khramtsova EA, Stranger BE. Assocplots: a Python package for static andinteractive visualization of multiple-group GWAS results. Bioinformatics.2017;33:432–4.

6. Haug-Baltzell A, Stephens SA, Davey S, Scheidegger CE, Lyons E. SynMap2and SynMap3D: web-based whole-genome synteny browsers.Bioinformatics. 2017;33:2197–8.

7. Usher W, Klacansky P, Federer F, Bremer P-T, Knoll A, Yarch J, Angelucci A,Pascucci V. A Virtual Reality Visualization Tool for Neuron Tracing. IEEE Trans.Vis. Comput. Graph. 2018;24:994–1003.

8. Zhang JF, Paciorkowski AR, Craig PA, Cui F. BioVR: a platform for virtualreality assisted biological data integration and visualization. BMCBioinformatics. 2019;20:78.

9. Balo AR, Wang M, Ernst OP. Accessible virtual reality of biomolecularstructural models using the Autodesk Molecule Viewer. Nat Methods. 2017;14:1122–3.

10. Goddard TD, Brilliant AA, Skillman TL, Vergenz S, Tyrwhitt-Drake J, Meng EC,Ferrin TE. Molecular Visualization on the Holodeck. J Mol Biol. 2018;430:3982–96.

11. Abel T. ReactJS: Become a Professional in Web App Development. USA:CreateSpace Independent Publishing Platform; 2016.

12. Gill A. AFrame: A Domain Specific Language for Virtual Reality: ExtendedAbstractAFrame: A Domain Specific Language for Virtual Reality: ExtendedAbstract. In: Proceedings of the 2Nd International Workshop on Real WorldDomain Specific Languages, vol. 4. New York: ACM; 2017. p. 1. –4:1.

13. Krzywinski M, Schein J, Birol I, Connors J, Gascoyne R, Horsman D, Jones SJ,Marra MA. Circos: an information aesthetic for comparative genomics.Genome Res. 2009;19:1639–45.

14. Pedersen BS, Yang IV, De S. CruzDB: software for annotation of genomicintervals with UCSC genome-browser database. Bioinformatics. 2013;29:3003–6.

15. GIANT consortium - Giant ConsortiumGIANT consortium - Giant Consortium[https://portals.broadinstitute.org/collaboration/giant/index.php/Main_Page].Accessed 17 December 2019.

16. Hunter DJ, Kraft P, Jacobs KB, Cox DG, Yeager M, Hankinson SE, WacholderS, Wang Z, Welch R, Hutchinson A, Wang J, Yu K, Chatterjee N, Orr N, WillettWC, Colditz GA, Ziegler RG, Berg CD, Buys SS, McCarty CA, Feigelson HS,Calle EE, Thun MJ, Hayes RB, Tucker M, Gerhard DS, Fraumeni JF Jr, HooverRN, Thomas G, Chanock SJ. A Genome-Wide Association Study IdentifiesAlleles in FGFR2 Associated with Risk of Sporadic Postmenopausal BreastCancer. Nat Genet. 2007;39(7):870-874.

17. Haiman CA, Chen GK, Vachon CM, Canzian F, Dunning A, Millikan RC, WangX, Ademuyiwa F, Ahmed S, Ambrosone CB, Baglietto L, Balleine R, BanderaEV, Beckmann MW, Berg CD, Bernstein L, Blomqvist C, Blot WJ, Brauch H,Buring JE, Carey LA, Carpenter JE, Chang-Claude J, Chanock SJ, Chasman DI,Clarke CL, Cox A, Cross SS, Deming SL, Diasio RB, Dimopoulos AM, DriverWR, Dünnebier T, Durcan L, Eccles D, Edlund CK, Ekici AB, Fasching PA,Feigelson HS, Flesch-Janys D, Fostira F, Försti A, Fountzilas G, Gerty AM,Giles GG, Godwin AK, Goodfellow P, Graham N, Greco D, Hamann U,Hankinson SE, Hartmann A, Hein R, Heinz J, Holbrook A, Hoover RN, Hu JJ,Hunter DJ, Ingles SA, Irwanto A, Ivanovich J, John EM, Johnson N, Jukkola-Vuorinen A, Kaaks R, Ko Y-D, Kolonel LN, Konstantopoulou I, Kosma V-M,Kulkarni S, Lambrechts D, Lee AM, Le Marchand L, Lesnick T, Liu J,Lindstrom S, Mannermaa A, Margolin S, Martin NG, Miron P, MontgomeryGW, Nevanlinna H, Nickels S, Nyante S, Olswold C, Palmer J, Pathak H,Pectasides D, Perou CM, Peto J, Pharoah PDP, Pooler LC, Press MF, Pylkäs K,Rebbeck TR, Rodriguez-Gil JL, Rosenberg L, Ross E, Rüdiger T, dos SantosSilva I, Sawyer E, Schmidt MK, Schulz-Wendtland R, Schumacher F, Severi G,

Sheng X, Signorello LB, Sinn H-P, Stevens KN, Southey MC, Tapper WJ,Tomlinson I, Hogervorst FBL, Wauters E, WeaverJE, Wildiers H, Winqvist R,Van Den Berg D, Wan P, Xia LY, Yannoukakos D, Zheng W, Ziegler RG,Siddiq A, Slager SL, Stram DO, Easton D, Kraft P, Henderson BE, Couch FJ. Acommon variant at the TERT-CLPTM1L locus is associated with estrogenreceptor–negative breast cancer. Nature Genetics. 2011;43(12):1210-1214

18. Sanciangco, Millicent D. (International Rice Research Institute); Alexandrov,Nickolai N. (International Rice Research Institute); Chebotarov, Dmytro(International Rice Research Institute); King, Ross D. (University ofManchester); Naredo, Ma. Elizabeth B. (International Rice Research Institute);Leung, Hei (International Rice Research Institute); Mansueto, Locedie(International Rice Research Institute); Mauleon, Ramil P. (International RiceResearch Institute); Orhobor, Oghenejokpeme I. (University of Manchester);McNally, Kenneth L. (International Rice Research Institute): Discovery ofgenomic variants associated with genebank historical traits for riceimprovement: SNP and indel data, phenotypic data, and GWAS result. 20182018.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.